Systematic review and meta-analysis of the efficacy of tirilazad in experimental stroke

Background and Purpose-Tirilazad is a candidate neuroprotective drug with reported efficacy in animal models of stroke that was, however, without benefit in clinical trials. This apparent contradiction might be explained if the animal studies were falsely positive, if the clinical trials were falsely negative, or if tirilazad was not tested under the same conditions in animal and clinical studies. Here we use systematic review and meta-analysis to describe the characteristics and limits to the neuroprotective action of tirilazad in animal models of stroke. Methods-Systematic review and meta-analysis of studies describing the efficacy of tirilazad in animal models of focal ischemia, in which outcome was measured as infarct volume and/or neurological score. Weighted mean difference random effects meta-analysis was used to measure overall efficacy in prespecified subgroups. Results-Eighteen studies describing outcome in 544 animals were identified. Study quality (median score, 5/10; interquartile range, 4 to 6) was similar to that seen in systematic reviews of other candidate neuroprotective drugs. Tirilazad reduced infarct volume by 29.2% (95% confidence interval 21.1% to 37.2%) and improved neurobehavioral score by 48.1% (95% confidence interval 29.3% to 66.9%). Conclusion-Tirilazad may have substantial efficacy in animal models of stroke, but this conclusion must be qualified because of the presence of potential sources of bias. (Stroke. 2007;38:388-394.)